The strength of associations between diet and chronic diseases is substantially underestimated due to the error involved in measuring diet. This error is composed of both random error and bias due to cultural, personal and other factors, and is present in both food frequency questionnaires (FFQ) and diet records. We propose a series of studies to improve diet assessment strategies aimed toward correcting the bias as well as random error. This promises to improve the epidemiologic studies and clinical studies using FFQs and diet records that have become the major methods for studying the prevention, management and treatment Of chronic diseases. Phase I (months 1-6) will include a baseline study to establish the direction and amount of bias associated with personal characteristics using doubly labeled water and biomarkers, and focus groups to identify the cultural, personal, and technical issues which affect women's ability to accurately report dietary intake. As part of the baseline study, subjects will complete standard questionnaires including the Marlowe-Crowne, Weinberger Adjustment Inventory, the Eating Disorder Inventory, the Restraint Scale and respond to the Stunkard's silhouettes regarding their body shape, their desired body shape and the range that represents healthy body shapes. Data on several other risk factors for chronic diseases will also be obtained including anthropometric measurements, family and personal history of chronic diseases such as osteoporosis, hypertension, diabetes, cancer and heart disease, cigarette smoking and weight history. In Phase II (months 7-25), we draw upon the focus group results and examine the relationships of various factors to women's reports of dietary intake including: personal variables listed above assessed via questionnaire; usual training vs. training directed toward the acceptability of reporting all foods, frequent eating, and large portion sizes; and written records vs. tape recorded. Phase III (months 26-33) will include three small studies examining variations in the physical layout of FFQs and portion size data bases. Phase IV (months 3442) will be a classic validity study, using the best methodologies from Phases I-III to evaluate how the "new improved" method compares with the original methods for obtaining diet intake data implemented in Phase I, and with doubly labeled water and plasma nutrients. Phase V (months 4348) will be the final analysis of the data and preparation of the final report.